Patients with inflammatory bowel disease (IBD) require long-term pharmacotherapy. Due to the wide variety of clinically available drugs, medication safety has become increasingly concerning. As an indispensable component of pharmaceutical formulations, drug excipients may induce various adverse reactions, including rash, allergic responses, and gastrointestinal discomfort, and may even lead to severe organic injury. However, they are often overlooked in clinical practice. Therefore, identifying and managing excipient-related adverse reactions is crucial for ensuring medication safety. To address this issue, this article systematically reviews common excipients in IBD therapeutics, the underlying mechanisms of excipient-induced adverse reactions, and corresponding management strategies, aiming to enhance clinicians' awareness of excipient-related issues, improve their ability to recognize and manage such reactions, and ultimately ensure patient safety.
ObjectiveTo explore the feasibility of using Large Language Models (LLMs) to generate evidence-based health science popularization materials by surveying public needs, designing specific prompts, and evaluating the quality of the generated content.MethodsAn online survey was conducted via SoJump platform to investigate public needs for healthy lifestyle information. The Kimi LLM was employed to generate health science popularization materials, with prompts optimized based on literature review and expert feedback. Two health educators independently evaluated the generated materials using the DISCERN instrument. Data were analyzed using SPSS 24.0 software.ResultsThe primary barriers hindering residents from accessing and using health information were distrust of publishing institutions (71.37%), excessive use of technical terminology (64.71%), and inaccessible channels (60.00%). The designed prompts, consisting of system and user prompts, comprised three sections: role definition, content generation, and structure/output formatting specifications. These prompts effectively guided the LLM to produce accessible health science materials with a standardized structure, including a title, introduction, specific interventions, and summary. Following prompt optimization, the DISCERN scores of the generated materials increased by (4.40±2.51) points, representing a statistically significant difference (P=0.041).ConclusionsHealth science popularization is a crucial factor in promoting the acquisition and utilization of health information among residents. Developing prompts to assist LLMs in creating health science materials holds significant potential; it facilitates health education work and enhances both the scientific accuracy and efficiency of content production.
Objective To develop an multimodal interpretable model that integrates whole slide images(WSI)and clinical features,and to validate its efficacy in predicting pathological complete response(pCR)in lung cancer patients following neoadjuvant therapy.Methods The clinicopathologic data who received neo-adjuvant therapy of patients as well as hematoxylin and eosin stained sections were retrospectively collected be-tween March 2015 and March 2025.For the WSI branch,the predictive performance of five pathology founda-tion models(CTransPath,Virchow2,H-optimus-0,Phikon-V2,and UNI-V2)was compared within the CLAM-SB framework to identify the optimal feature extractor.The clinical branch was constructed using the ex-treme gradient boosting(XGBoost)model.Subsequently,a multimodal prediction model for pathologic com-plete response(MP-pCR)was established through decision-layer logistic regression fusion.Model performance was evaluated using the area under the receiver operating characteristic curve(AUC),accuracy,sensitivity,specificity,F1-score,and Brier score.Interpretability analysis was performed using attention heatmaps and the Shapley Additive Explanations(SHAP)algorithm.Results A total of 728 patients were enrolled in this study.Among them,a total of 676 patients from the Fourth Hospital of Hebei Medical University were selected as internal dataset,which was randomly divided into training set(n=536),validation set(n=70),and in-ternal test set(n=70)at a ratio of 8∶1∶1.Additionally,32 patients from the Affiliated Hospital of Hebei University and 20 from Handan First Hospital were selected as external validation set 1 and external validation set 2,respectively.In the internal testing set,UNI-V2 emerged as the top-performing feature extractor for the WSI branch,achieving an AUC of 0.774(95%CI:0.688-0.861).The resulting MP-pCR multimodal model yielded an AUC of 0.812(95%CI:0.725-0.899)in the internal testing set,outperforming both the WSI-only model(0.774)and the clinical-only model(0.780),with an accuracy of 81.8%,sensitivity of 70.6%,specificity of 86.5%,and a Brier score of 0.125.In external validation set 1,the MP-pCR model achieved an AUC of 0.746(95%CI:0.598-0.898)and an accuracy of 75.0%;in external validation set 2,it achieved an AUC of 0.722(95%CI:0.538-0.918)and an accuracy of 80.0%.Attention heatmaps re-vealed that the model-focused regions were primarily concentrated within the tumor parenchyma and treatment-related peritumoral areas.SHAP analysis indicated that preoperative treatment regimen,pathological diagnosis,tumor-to-stroma ratio,age,and smoking history were the top five contributors to pCR prediction.Conclusions MP-pCR model developed and validated based on multicenter data outperforms single-modality models in pre-dicting pCR for lung cancer.The interpretability results demonstrate good consistency with clinicopathologic knowledge,suggesting that the model holds promise as a potential auxiliary reference for the clinical evaluation of treatment response and the optimization of individualized therapeutic decisions.
ObjectiveBased on the "gene-brain-cognition" model, this study aimed to investigate the association between the rs4860671 polymorphism of the EPHA5 gene and attention-deficit/hyperactivity disorder (ADHD), and to analyze its impact on executive function and resting-state brain activity in affected children.MethodsThis was a case-control study. Children with ADHD attending the outpatient clinic of Peking University Sixth Hospital from 2015 to 2020 and typically developing children from general primary and secondary schools in Beijing during the same period were enrolled as subjects. Genotypes of the EPHA5 rs4860671 polymorphism were determined in both groups. Resting-state functional magnetic resonance imaging was used to assess spontaneous brain activity. In addition, the ADHD Rating Scale and the Behavior Rating Inventory of Executive Function were administered to evaluate ADHD core symptoms and executive function, respectively. Genetic association analysis, analysis of covariance, and mediation analysis were systematically performed to explore the associations of the rs4860671 polymorphism with ADHD clinical symptoms, executive function, and brain functional activity.ResultsA total of 244 children with ADHD and 83 healthy controls were included. Among them, 188 children with ADHD and 70 healthy controls completed MRI scanning. Allelic analysis revealed that the frequency of the G allele of rs4860671 was significantly higher in children with ADHD than in healthy controls (77.5% vs. 69.9%, P=0.049). Genotype distribution analysis also demonstrated a significant difference between the two groups under the dominant model (P=0.031). Regarding executive function, children carrying the GG genotype of rs4860671 had higher scores in emotion control compared with those carrying the A allele (AA+AG) (15.99±4.62 vs. 14.53±4.03, P=0.017), indicating poorer emotion regulation ability. Further mediation analysis showed that emotion control fully mediated the relationship between rs4860671 genotype and core symptoms of ADHD. Neuroimaging analysis revealed that among children with ADHD, carriers of the GG genotype exhibited decreased spontaneous brain activity in the right inferior occipital gyrus and left middle occipital gyrus compared with A allele carriers.ConclusionsThe rs4860671 polymorphism of the EPHA5 gene may be involved in the pathogenesis of ADHD by influencing emotion regulation ability and visual network functional activity. This finding provides new scientific evidence for a deeper understanding of the neurobiological mechanisms underlying ADHD.
In accordance with the Pharmaceutical Administration Law of the People's Republic of China revised in 2019 and the Implementing Regulations of the Pharmaceutical Administration Law of the People's Republic of China effective in 2026, Peking Union Medical College Hospital led the development of The Chinese Management Consensus on Compassionate Use of Investigational New Drugs in Medical Institutions(2026). This consensus was formulated by convening a multidisciplinary panel of experts, including health policy specialists, clinicians, pharmacists, methodologists, ethicists, and industry representatives, following the consensus development process recommended by the World Health Organization handbook and the Chinese Medical Association. As the first management consensus in China specifically addressing compassionate use of investigational drugs, this document provides 19 recommendations across eight key domains: eligibility criteria for compassionate use, scope of application, target patient populations, responsibilities of relevant stakeholders, key points for application and review, patient informed consent, risk prevention and control, and cost-bearing arrangements. This consensus aims to standardize the clinical practice of compassionate use of investigational drugs within medical institutions, enhance awareness of this issue among healthcare institutions in China, address the clinical needs of critically ill patients requiring urgent treatment, and provide recommendations and references for the establishment of standardized compassionate use management procedures.
The crosstalk among pyroptosis, apoptosis, and necroptosis has facilitated the emergence of the innovative concept of PANoptosis. As a form of immunogenic cell death, PANoptosis is closely associated with various tumors through its unique regulatory mechanisms, thereby offering novel directions and insights for tumor immunotherapy. This review summarizes the mechanistic underpinnings of PANoptosis and the current evidence regarding its role in tumor progression. Furthermore, we discuss immunotherapeutic strategies targeting PANoptosis, with the aim of transitioning the research paradigm of tumor cell death from a unimodal to a multidimensional regulatory framework, and providing new perspectives for the clinical management of malignancies.
ObjectiveTo evaluate and project the disease burden of early-onset colorectal cancer (EOCRC), and to identify its major risk factors.MethodsData from the Global Burden of Disease Study 2021 were used. EOCRC was defined as colorectal cancer diagnosed at ages 15-49 years. We extracted the absolute numbers and crude rates of incidence, mortality, and disability-adjusted life years (DALYs) of EOCRC in the world, China, and the United States from 1990 to 2021. Age-standardized rates were then calculated accordingly, and the associated risk factors were summarized. Joinpoint regression was applied to assess temporal trends in global EOCRC age-standardized incidence rate (ASIR), mortality rate (ASMR), and DALY rate. The Bayesian age-period-cohort (BAPC) model was used to project the EOCRC burden from 2022 to 2040.ResultsIn 2021, an estimated 211 900 new EOCRC cases were reported globally, corresponding to an ASIR of 5.37 per 100 000, along with 79 500 deaths and a total of 4.00 million DALYs lost. China experienced a substantially higher burden than the global average, with an ASIR of 10.02 per 100 000, an ASMR of 3.11 per 100 000, and an age-standardized DALYs rate(ASDR) of 160.93 per 100 000, disproportionately affecting males. The United States had a higher ASIR(10.73 per 100 000) than China, but exhibited lower ASMR(2.53 per 100 000) and ASDR(127.81 per 100 000). From 1990 to 2021, ASIRs increased globally (AAPC=0.374) and more markedly in China (AAPC=1.423), whereas ASMR (AAPC=-0.855) and DALY rates (AAPC=-0.834) declined. In China, ASMR and DALY rates declined overall but rebounded in recent years, whereas in the United States, ASIR began to decrease after 2017, while ASMR and DALY rates remained relatively stable. By sex, China's male ASIR increased by 89.7% between 1990 and 2021, with rises also observed in ASMR and DALY rates in recent years. Regarding risk factors, insufficient whole-grain intake and excessive red meat consumption emerged as the leading modifiable contributors to EOCRC mortality and DALYs both globally and in the two countries, jointly accounting for 31.20% and 30.82% of the global burden, respectively. Heavy alcohol consumption and tobacco use were particularly significant contributors among Chinese men. Projections indicated that by 2040, China's ASIR is projected to rise to 17.04 per 100 000, indicating a continued increase in disease burden, whereas the United States is expected to maintain a downward trend.ConclusionsAlthough China's EOCRC mortality and DALY burden have declined over the past three decades, the incidence remains markedly higher and is rising more rapidly compared with the global level and the U.S. More effective public health strategies are urgently needed, including promoting earlier and wider implementation of colorectal cancer screening, strengthening follow-up and referral systems, improving dietary structure, and intensifying tobacco and alcohol control, to mitigate the growing burden of EOCRC.
The endochondral ossification of the growth plate is crucial for the longitudinal growth of long bones. Mechanical loading plays an important regulatory role in endochondral ossification. Absence or improper mechanical loading can inhibit growth plate development, while moderate mechanical stimuli promote chondrocyte proliferation and differentiation, thereby facilitating long bone growth. Mechanosensitive structures, such as primary cilia, mechanosensitive ion channels, integrins, and the cytoskeleton, play key roles in the regulation of mechanical loading. This review summarizes how mechanical loading influences growth plate endochondral ossification and discusses potential underlying mechanisms, offering theoretical insights for the management and research of bone development disorders in children and adolescents.
Polycystic ovary syndrome (PCOS) is a complex and heterogeneous endocrine disorder affecting 5%-15% of women of reproductive age worldwide. It is characterized by hyperandrogenemia, ovulatory dysfunction, and polycystic ovarian morphology, and is associated with long-term risks of chronic conditions such as obesity, cardiovascular disease, and diabetes mellitus. Although the etiology of PCOS remains incompletely understood, the intricate interplay of genetic predisposition, environmental factors, and epigenetic modifications plays a pivotal role in its pathogenesis. ETS1 (E26 transformation specific 1), a transcription factor, is involved in regulating diverse critical biological processes, including cell proliferation, differentiation, angiogenesis, immune response, and apoptosis. Accumulating evidence suggests that ETS1 may play a significant role in the development and progression of PCOS, particularly in ovarian function, hyperandrogenesis, and inflammatory responses. This review aims to comprehensively summarize recent advances in understanding the role of ETS1 in the pathophysiological mechanisms underlying PCOS, elucidate its potential molecular mechanisms, and evaluate its potential as a therapeutic target for PCOS. The findings may provide insights into the etiology of PCOS and facilitate the development of novel therapeutic strategies.
As a major challenge in the global field of critical care medicine, sepsis is associated with high morbidity and mortality, driving intensive research into the development of precise risk prediction models. Among these, machine learning and deep learning-based predictive models have garnered particular attention. This article systematically reviews the evolution of sepsis risk prediction models and the challenges associated with their clinical translation, with a focus on their clinical applicability in common septic complications, including acute kidney injury, encephalopathy, and coagulation disorders. Current evidence indicates that machine learning models, by integrating multidimensional dynamic data from electronic health records, significantly enhance early warning capabilities and individualized predictive performance. However, major bottlenecks limiting clinical translation include insufficient data standardization, lack of model interpretability, and inadequate external validation. Future research should prioritize multicenter collaboration, time-series modeling strategies, and explainable artificial intelligence frameworks to optimize clinical translation pathways and facilitate the paradigm shift from traditional empirical approaches to data-driven precision strategies in sepsis risk stratification.
Medical imaging foundation models, as an important direction in the intelligent analysis of medical images, are driving a paradigm shift in this field—from task-specific modeling for individual tasks toward a new paradigm grounded in large-scale pre-training and general-purpose representation learning. These models have demonstrated considerable potential in multimodal information integration, cross-task transfer, multi-scenario adaptation, and generative interaction. However, most current studies remain based on benchmark datasets, retrospective cohorts, or controlled experimental conditions, and still face challenges such as insufficient data representativeness, task settings that deviate from real-world clinical workflows, inadequate validation of generalizability and robustness, and ambiguous boundaries of ethical oversight and accountability. This article reviews the research background, key technologies, and advances in the applications of medical imaging foundation models, and further discusses practical challenges including data governance, workflow integration, and ethical regulation, with the aim of providing guidance for related research and clinical translation.
Factorial design is an important method for optimizing complex strategies in implementation science. This paper systematically elucidates the origin, types, and core concepts of the factorial design, and then use a brief smoking cessation intervention project as an example to illustrate its specific applications in identifying key factors and testing interactions. Factorial design has the following typical advantages: it can test synergistic and antagonistic effects between factors, provide effect estimates that are broadly valid across the experimental conditions, and offer high statistical efficiency. At the same time, it also has certain limitations, including that its complexity is positively correlated with the number of factors, the need for large sample sizes to detect interactions, the potential for insufficient acceptability of some strategy combinations due to ethical concerns or participant burden, and the complexity of the analytical methods required. In future research, its application should be further optimized by appropriately controlling the number of factors, clearly defining research objectives, and evaluating intervention acceptability, in order to promote the development of precision implementation science.
Drug-target interaction (DTI) is fundamental to novel drug research and development (R &D), playing a critical role in elucidating drug mechanisms of action and improving the cost-effectiveness of drug discovery. Although China has seen rapid accumulation of drug-target resources in recent years, the industry still faces challenges such as a shortage of original targets and excessively high target concentration. The rapid advancement of artificial intelligence (AI) has provided an efficient technological pathway for DTI prediction, establishing it as a core tool for accelerating drug discovery. This paper systematically reviews the fundamental data support framework for DTI prediction tasks, elaborating in detail on three core data types— drug characterization, target characterization, and drug-target associations-as well as the classification and application boundaries of mainstream benchmark datasets tailored to different tasks within the field. On this basis, it comprehensively surveys the technological evolution of AI-driven DTI prediction, summarizing the core paradigms and technical advances of traditional machine learning and deep learning methods in a single-modality setting, alongside cutting-edge multimodal approaches that integrate two modalities (cross-subject/within-subject fusion) and three or more modalities for multidimensional information integration. Finally, this paper analyzes the current major challenges in data quality, model generalizability and interpretability, and real-world deployment, while also discussing future trends in standardized dataset construction, generative AI, and large-scale biomedical foundation models, aiming to provide a reference for both research and industrial applications in the AI-driven DTI prediction field.
Pharmacist training in the pharmacy intravenous admixture service (PIVAS) center is a core component in ensuring the quality of finished infusions and medication safety. Based on the Quality Management Standard for Centralized Intravenous Admixture (2010) and the Notice on Issuing the Guidelines for the Construction and Management of Pharmacy Intravenous Admixture Services (Trial) (2021), the PIVAS center of Peking Union Medical College Hospital, guided by the concept of post competency, has developed a tiered training model tailored to the work requirements of different PIVAS positions (such as prescription review, drug admixture, and finished product verification). Concurrently, by integrating common admixture errors observed during training and routine practice, the program incorporates error-prone content into teaching priorities and dynamically adjusts the training plan, thereby providing a reference for the development of pharmacist training systems in China.
ObjectiveTo explore the clinical value of a multimodal predictive model based on multiparametric magnetic resonance imaging(MRI) radiomics combined with deep learning(DL) features for the preoperative noninvasive assessment of mismatch repair-deficient(MMRd) status in endometrial cancer(EC).MethodsPatients diagnosed with EC at Peking Union Medical College Hospital from January 2015 to December 2021 were retrospectively enrolled and randomly divided into a training set and a validation set at a ratio of 8∶2. Relevant clinical data were collected, and radiomics features and DL features were extracted from preoperative contrast-enhanced T1-weighted imaging(CE-T1WI), fat-suppressed T2-weighted imaging(fs-T2WI), and diffusion-weighted imaging(DWI) sequences. High-dimensional feature selection and dimensionality reduction were performed sequentially using the recursive feature elimination(RFE) algorithm to generate a radiomics score(Rad-score) and a deep learning score(DL-score), respectively. Multivariate logistic regression was utilized to construct a clinical model, a pure radiomics model, a clinical-radiomics model, and an integrated multimodal model incorporating clinical indicators, Rad-score, and DL-score. Model performance was assessed and compared using area under receiver operating characteristic curve(AUC) and DeLong test.ResultsA total of 509 patients were enrolled in this study, comprising 413 in the training cohort and 96 in the validation cohort. Independent predictors: Multivariate analysis indicated that preoperative fasting blood glucose level, histological grade, lymph node metastasis status, Rad-score, and DL-score were all independent significant predictors of MMRd status in EC patients. The integrated multimodal model demonstrated optimal predictive performance with an AUC of 0.699(95% CI: 0.635-0.763) in the training set, which was superior to the clinical model(AUC=0.629, 95% CI: 0.561-0.697) and the pure radiomics model(AUC=0.641, 95% CI: 0.575-0.706). In the validation set, the integrated model maintained good generalizability, achieving an AUC of 0.655(95% CI: 0.535-0.775), and its diagnostic efficacy was higher than that of the clinical model(AUC=0.578, 95% CI: 0.450-0.705) and the pure radiomics model(AUC=0.611, 95% CI: 0.488-0.734). According to the DeLong test, the incorporation of DL features resulted in the clinicalradiomicsdeep learning model performing better than both the clinicalonly model(P=0.027) and the radiomicsonly model(P=0.044) in the training cohort.ConclusionsThe initially developed clinical-radiomics-deep learning model exhibits a certain predictive potential for the MMRd status in patients with EC. The inclusion of DL features may help complement the limitations of traditional evaluations, offering a preliminary radiological reference for preoperative non-invasive screening. However, given the current diagnostic performance, its overall accuracy and clinical generalizability warrant further validation in multi-center, large-sample external cohort studies.
As pilot programs for hospice and palliative care deepen in China, conflicts have emerged in clinical practice between Western ethical principles centered on "patient autonomy" and the traditional Chinese family-centered decision-making model, leading to widespread dilemmas regarding informed consent and family conflicts in decision-making. The sinicization of palliative care practice should not involve simply replacing one model with another. Instead, it should be based on a profound understanding of the local culture to achieve "relational autonomy", thereby constructing a palliative care ethical system that genuinely aligns with China's national context and can effectively guide clinical practice.
Objective To analyze global trends and research hotspots in exercise intervention for over-weight and obesity from 2010 to 2024,to provide novel perspectives for comprehensive management research on overweight and obesity.Methods Relevant literature published between January 1,2010,and December 31,2024,was retrieved from all sub-databases of the Web of Science Core Collection,including articles and reviews.CiteSpace 6.4.R1 was used to perform co-occurrence and burst analysis of journals,institutions,au-thors and keywords.VOSviewer 1.6.20 was applied to construct co-citation networks among countries/regions and references.SPSS 26.0 was utilized to analyze trends in publication volume and citation frequency.Results A total of 113 080 publications on exercise interventions for overweight and obesity were obtained,including 98 188 articles and 14 892 reviews.Both annual publications(t=8.75,P＜0.01)and citation frequency(t=8.14,P＜0.01)showed significant growth trends.The United States contributed the highest number of publications(40 711)and citations(1 513 360).England contributed the highest Hirsch index(H-index)(246).Harvard University contributed the highest number of publications(8290),citations(503 061),the average citations per paper(60.68)and H-index(176).Moreno L A,from the University of Zaragoza in Spain,had the highest number of publications(160).High-frequency keywords primarily included"physical activity""obesity""overweight""body mass index",and"exercise".Burst keywords mainly comprised"cardiometabolic risk""eating habits",and"high-intensity interval training".Conclusions Over the past 15 years,the field of exercise intervention for overweight and obesity has received considerable academic atten-tion.Key research hotspots focus on monitoring the incidence of overweight and obesity and the correlation be-tween physical activity and the health outcomes of individuals with overweight and obesity.Future research di-rections emphasize the effect of high-intensity interval training on the cardiometabolic health of individuals with overweight and obesity,the interaction of dietary habits and physical activity,and the focus on enhance physi-cal activity levels are the primary global trends in this research field.
ObjectiveTo develop a deep learning model based on arteriovenous dual-phase CT imaging features and evaluate its diagnostic value in differentiating benign from malignant pancreatic cystic lesions(PCLs).MethodsPreoperative contrast-enhanced CT images of patients with histopathologically confirmed PCLs at Peking Union Medical College Hospital from June 1, 2014 to May 31, 2023 were retrospectively collected. The data were randomly partitioned at the lesion level into training, validation, and test sets in a 3∶1∶1 ratio. CT images were preprocessed and regions of interest were delineated. Using postoperative pathological results as the reference standard, five deep learning models(ResNet50, DenseNet121, ResNeXt50, EfficientNet-b5, and MobileNetV2) were constructed to extract arteriovenous dual-phase CT imaging features for binary classification of PCLs. The optimal model was selected based on the area under the curve(AUC), accuracy, sensitivity, and specificity. Further comparative analyses were conducted against single-phase CT-based deep learning models, conventional radiomics models, and radiologist interpretations to comprehensively evaluate the performance of the arteriovenous dual-phase CT-based deep learning model in the differential diagnosis of PCLs.ResultsA total of 480 patients with 485 lesions(206 malignant and 279 benign) were ultimately enrolled. The training, validation, and test sets comprised 291, 97, and 97 lesions, corresponding to 288, 96, and 96 patients, respectively. In the validation set, the ResNeXt50 model based on arteriovenous dual-phase CT features achieved an AUC of 0.837(95% CI: 0.748-0.915), an accuracy of 77.32%(95% CI: 67.70%-85.21%), a sensitivity of 80.49%(95% CI: 65.13%-91.18%), and a specificity of 75.00%(95% CI: 61.63%-85.61%). In the test set, the corresponding values were 0.822(95% CI: 0.737-0.904), 73.20%(95% CI: 63.24%-81.68%), 82.93%(95% CI: 67.94%-92.85%), and 66.07%(95% CI: 52.19%-78.19%), demonstrating overall superior performance. Calibration curves indicated that the predicted probabilities of the model were generally consistent with observed outcomes, albeit with certain shortcomings in probability calibration. Decision curve analysis demonstrated that, overall, the use of this model for clinical decision-making conferred a net benefit to patients. Compared with single-phase CT-based models and conventional radiomics models, the ResNeXt50 model incorporating arteriovenous dual-phase CT features exhibited superior overall performance, and its diagnostic performance was comparable to that of radiologists, with good consistency.ConclusionsThe deep learning model based on arteriovenous dual-phase CT imaging features demonstrates diagnostic value in differentiating benign from malignant PCLs and may serve as an adjunctive reference for preoperative clinical assessment. However, its stability and generalizability warrant further validation in larger cohorts and with external datasets.
Pathology diagnosis is the "gold standard" for clinical disease diagnosis and treatment. Intelligent pathology image analysis has significant clinical value in improving diagnostic efficiency and ensuring consistency. This paper systematically reviews the core multiple instance learning paradigm for whole slide image analysis, and elaborates the pretraining system, core functional branches and latest advances of pathology foundation models. Taking clinical usability as the core evaluation dimension, it establishes a multi-dimensional evaluation framework, analyzes the clinical translation potential of different technical routes, and points out the core contradiction of "excellent laboratory performance but insufficient clinical usability" in current pathology foundation models. It further dissects the key clinical translation bottlenecks of pathology foundation models, outlines future application and development directions, and provides a reference for technological research and development as well as clinical translation in computational pathology.
Clinical practice guidelines serve as a crucial bridge linking research evidence, clinical practice, and health policy decision-making. However, the guideline domain continues to face persistent challenges, including duplicate guideline development, variable methodological quality, lack of transparent reporting, insufficient implementation and translation, and delayed updates. Focusing solely on "how to develop a high-quality guideline" is no longer sufficient to meet the modern healthcare system's demands for timely, trustworthy, actionable, and dynamically updated recommendations. The emergence of the guideline ecosystem marks a shift in guideline research and practice from a linear development process toward a more systematic governance model. Rooted in the evidence ecosystem, the guideline ecosystem has its own distinct functional positioning, encompassing recommendation formulation, clinical implementation, and life-cycle updating. Looking ahead, China should leverage guideline registries, integrated quality assessment systems, implementation science, dynamic updating mechanisms, and artificial intelligence technologies to promote the transformation of its guideline system from scale expansion to an ecosystem-based model characterized by high quality and sustainability.
